AI has gone very quickly from niche to mainstream in early-childhood education. With a market worth at least ten billion yuan, Jiemian News talks to parents and entrepreneurs.
Photo from Unsplash.
By ZHA Qinjun
Earlier this year, a promotion by ed-tech company Zebra AI caught the eye of ZHANG Na, a working mom in Wuhan. For just 99 yuan (less than $15) her four-year-old daughter could enjoy 10 AI-powered English lessons and a gift box chock full of textbooks and learning toys. Zhang didn’t hesitate for long: “The gift box itself is worth more than 99 yuan!”
Zebra AI is one of many arriviste tech businesses storming early-childhood education. Lessons rely on multimedia content and interactive features rather than human teachers. They are considered a good fit for preschool foreign language learning, thanks to the well-established curriculum and relatively low algorithmic requirement for natural language processing. It is the latest battleground in the fierce online education market and seen as a gateway for a wider range of educational products.
AI in early-childhood education has gone from niche to mainstream very quickly.
“The arena has attracted a lot of investment since last year,” said LI Yifei, director of Edbeta Fund, a venture capital firm specializing in education. As tech giants and start-ups scramble to launch AI lessons for preschoolers, she believes players are still operating without a distinguishable business model.
Zebra AI is an arm of ed-tech unicorn Yuanfudao, has done well since it was launched three years ago, reaching a million long-term users in July, and reportedly generating 300 million yuan in revenue.
It is not the first company to take Artificial intelligence to the early-childhood market. Jiliguala, an English-learning app for children under nine years old, has amassed 30 million users since 2014. Although the concept was new to most parents, low prices quickly cornered the market in small cities and rural areas, where half their users live.
Consultancy iFenxi estimates the market to be worth at least ten billion yuan. AI classes generate more revenue per customer and are believed to “deliver results” easily for both customers and the company.
“Millennial parents are keen to spend on early-childhood education,” said HUANG He, CEO of Palfish whose lines include live lessons and interactive picture books. Palfish has recently launched an AI product.
Such potential has attracted investors, many backed by tech giants such as ByteDance, Tencent, and TAL, a Nasdaq listed tech education company. Web traffic is key to early success and ensuing growth.
A tiny percentage steered from the existing hundred million user bases of Douyin (by ByteDance) or WeChat (by Tencent) would kickstart any education line. Parents who buy preschool lessons are highly likely to spend more in the years to come.
Despite the dazzling offerings, most AI learning tools are still far from being truly “intelligent.”
Developing an AI language lesson requires advanced machine learning and voice recognition technologies compares to in-person learning, as well as deep knowledge of linguistics and education, according to Toby Mather, CEO of Linguimi, a London-based English education start-up that launched in China in 2019. Linguimi has invested heavily in AI since day one due to “a lack of good human teachers.” During its market research, Linguimi found many language learning chatbots were not responsive enough to create a truly interactive learning experience.
“Even a toddler can easily tell the difference between a real human teacher and a robot,” said Mather.
On the other hand, many AI lessons load up too much cartoon content - as all else fails to capture the attention of the kids - which according to TANG Zhenhua, CEO of online early education institution NPY Mathematical Thinking, is “passive” and “not creative.”
Other so-called AI lessons have human teachers, who provide small tips and feedbacks in online group chats. Parents are not impressed. “It’s hard to find information in a large group chat,” said one. “The teacher doesn’t have the time or energy to give feedback to each student”
Without any original or interactive content, few products are pulling up many trees. Huang admits that AI education companies offer “similar products targeting the same user base.”
“AI lessons have only skimmed the surface,” said Edbeta’s Li.
Before technology and content barriers go up, there is very little choice between companies except for pricing and marketing gimmicks. Some, Zebra AI, for example, rely on discounts and giveaways. Lingumi has a subscription model that offers three levels of membership that can be shared by children in the same household. A few other companies have expanded their offerings to teach preschoolers writing and critical thinking.
Mather describes the current business model as “volume reliant.” The upfront costs in technology and content are covered only as the user base grows. “It’s like making a series on Netflix,” he said. “It loses money in the early years but with a larger volume of users, the cost will be covered.”
Lingumi’s subscription model is based on feedback from children and parents but Mather doesn’t rule out taking on corporate clients in the future.
The rich tech giants, on the other hand, are preparing for an all-out advertising war. Tencent reportedly has a billion yuan to splash on ABCmouse, an English learning app for children between 3 to 8. Similarly, English learning app Guagualong, backed by ByteDance, has budgeted 2 billion yuan for advertising. Zebra is reported to have spent at least 1.2 million yuan recently on a 15-minute live streaming session hosted by LUO Yonghao, a once-famous English teacher turned into an infamous celebrity entrepreneur who ran into a 600 million yuan debt after failed in his smartphone business.
The wearying price wars and endless spending sprees are transforming the market. The average cost per lead, or the amount of money needed to generate one customer inquiry, increased fivefold to 1000 yuan from last year’s 200 yuan, while the average price for a trial lesson dropped from 89 yuan to 49 yuan.
“Only tech giants or unicorns can afford to bleed money like this,” Li said. “There’s barely any breathing space for the little guy.” She thinks that the money-losing game is expected to continue until the industry begins to consolidate, then industry leaders will begin to turn a profit.
This may not come as early as many hope. It will be a struggle to retain customers when deep-discount trials end.
When Zhang used up her ten-lesson offer from Zebra AI, she balked at the 2600 yuan annual fee: “AI might be all the rage, but I’m not convinced it’s a good way of teaching.”
